Artificial intelligence is a sprawling, fascinating subject that marketers are working hard to comprehend, even as it advances rapidly. And Microsoft’s search engine, Bing, is positioned at the centre of these advances. As a challenger brand to the obvious giant (in the UK, one in four search queries go through Bing’s platform), they’re focused on getting the most out of AI - the fuel that powers search engines - so they can continue growing at their current pace.
Microsoft’s focus on this technology may be most apparent to the advertising community in its partnership with Publicis building Marcel, the Groupe’s much-discussed AI platform. It also offers a range of AI tools and frameworks that marketers are using in strategic and creative ways. And of course it also has Cortana - its virtual assistant, which grew out of developments made in the field of search.
LBB’s Alex Reeves asked Axel Steinman, VP strategic search for Microsoft Bing, to share some insights from the world of search - a relevant outsider’s perspective for the marketing world.
AI is the foundation of search
“At the core of Bing is artificial intelligence. And at the same time Bing became the core of the artificial intelligence efforts of Microsoft. It started all in Bing because it’s about understanding artificial intelligence in terms of managing a huge amount of data to really try to extract consumer understanding, consumer insights, with the purpose of understanding intent.
When people are using their digital devices, sometimes they are trying to entertain themselves watching a video or sharing pictures with a million friends. But most of the time people are trying to accomplish something. They have a task they want to complete.
In all those different instances, what is the true intent from people? You need machine learning.
For instance, if you are walking down the street with your mobile device and you write or say ‘sushi’, it’s probably likely that you’re looking for the closest restaurant at walking distance. If you are in the same situation but driving a car, maybe the same, but you are willing to go a few miles because you’re driving. If you’re on a PC in your office or home, you might either be looking for delivery of sushi or maybe recipes. We need to understand that intent to serve the best option.
Artificial intelligence is the core engine behind a search engine because there is no other way to use search without artificial intelligence. You get phrases, with voice search you get long phrases. You need to understand natural language and based on history, based on trending topics, multiple different things, really understand what the consumer is trying to do.”
Search is a less invasive space for marketers to meet consumers
“The problem with digital advertising in general is that it has played with the volumes more than with the real consumer value and impact. When you look at display, branding, programmatic, those are places that use volumes at the risk of hurting brands.
Sales says ‘I need 10,000 leads to sell this new big product. Marketing, here is your budget. Get me 10,000 clicks.’ So marketing goes and buys 10 million impressions at the price of the budget. 10 million impressions at 0.1% clickthrough rate gets the 10,000 leads. Sales is happy, marketing is happy. Everyone is happy. And no one is thinking ‘what happened with the 99.9% of users that did not click and probably now hate your brand because they came in in an invasive, interruptive way?’ That is a problem that you don’t face in search because search is where people express the most clear intent.
The interesting thing is because of this overwhelming media and people becoming more impervious to ads, and you see this self-inflicted problem in the industry with ad blockers and things. We are seeing now that people are using search all along the consumer decision journey. They are using search when they research, when they want to be aware of products, when they want to compare, all the way to the moment in which they make the last click.
So for marketers and advertisers, search is becoming more relevant than ever before as the most impactful way of touching the consumer in the right mindset - when they are expressing clear intent.”
AI is providing understanding of search audiences
“We are providing insights to our clients on the different behaviours of the audience. Of course statistical trends, not individualised data. We are building audiences within search, which is an innovation. Google doesn’t offer that. We are offering audience group targeting, precisely because of understanding elements along the consumer decision journey and how marketers can jump in there.
Then as we develop audiences and new insights from this intelligent search, we will be able to provide marketers with even richer information. It’s a constant evolution. That is a key element. Today everything is about data.
Marketers that are really exploiting a specific feature we are offering that means they can find, in search, consumers that have been on their sites. It’s first party data, so consumers are in a sense connected with that brand. They are smart using that because it’s a consumer that’s more advanced in the decision journey, closer to you, so [the brand is] more interested in that click.
That’s a reality. What do you understand from those consumers is how to touch them at the moment when the mindset is right - when they have intent - not when they don’t want to be bothered. We don’t believe in ruining consumer experiences to drive one more pound today. We are seeing consequences in the world for some companies because of focusing that way. We prefer not to.”
Chatbots are easy to build now
“Another element that connects with search but also with artificial intelligence is that a lot of clients are starting to use chatbots. We have a number of artificial intelligence services available to our clients. If you go to Microsoft Cognitive Services, there are services you can use to implement in your own website.
The Microsoft Bot Framework
allows you to develop chatbots very quickly. Clients are starting to map their search journey with their chatbox services, with their display and offline campaigns, connecting everything together. I think that’s smart use of search.
We have a restaurant in Seattle that has a chatbot now. When they are closed you can still ask the chatbot about parking, their menu, vegetarian options, whether they are kid friendly. It’s a tiny restaurant - a ‘moms and pops’ business. They couldn’t afford a call centre, people on the phone. It didn’t exist as an opportunity for them before.”
AI is even more key when it comes to voice search
“Voice search is another buzzword. It’s coming, clearly. It’s real. But how to use it is what people are still a bit lost on. I was recently listening to a panel - all smart people, advertisers and brands. It was interesting because the way they were articulating voice search crystallised in my mind that people don’t know what to do with it.
Voice search is queries logged on the system, except they are initiated by voice. It goes to the same system, creates queries based on keywords. Except the queries come from verbalised long phrases. So the first difficulty there needs artificial intelligence - to understand natural language - because we speak several times longer than we write. So all the nuances around talking to a device from a technology perspective has to be [understood]. In the past it was awful. It’s improving quickly as the system gains more data etc.”
How marketers can engage best with voice search
Scenario 1 - Simple tasks
“The first scenario is a task completion purpose. So a command to give to your device by voice. ‘What’s the temperature?’ ‘Turn on the lights’ or ‘Set a reminder for 3pm to meet Alex.’ I don’t see any marketing or advertising options there. It’s a great benefit for consumers, will create a relationship with the device of a different type and will be used in some cases. But there are nuances of etiquette and things like that.”
Scenario 2 - Complex tasks and added screens
“The second is what you really can call voice search. Which is a more complex task that you need to accomplish, where you need options, detailed information. And I don’t see that scenario playing out well screenless. The concept of talking instead of writing makes sense. We speak three times faster than we write. It’s more comfortable. But the responses need a screen. So I think there’s a combined scenario there of voice and screen.
That opens a new spectrum of ways of interacting with devices - the way of responding to that that might combine voice and visuals. And this is important because we are have three referential systems - the way we capture and process the input from the environment - visual, auditory and kinaesthetic. Most people have a preferred one. And more than half of people have visual as their preferred one. So what it means is if I ask you ‘give me ten options for people to meet at this event’ if you write and show me a visual list I might remember seven because I am visual. If you tell me ten, I will remember two or three. So voice search responding without a screen, I honestly don’t think it’s an impactful scenario.”
Scenario 3 - Content delivery
“The third scenario struck me on this panel because they were talking about it all the time - they were talking about content. You command Spotify to play some music or the BBC to play a podcast. You command the receiving of content. That’s called radio on demand, with artificial intelligence and machine learning to understand your tastes and propose ideas. That’s it. What’s the innovation in that, except that you talk to the radio instead of moving the dial?
Let’s be clear. It’s great. But it’s just radio. And the marketing opportunities are the same. It’s been around for 100 years and it will never die because there is always some time when we cannot use our eyes. We are driving and need to pay attention and things like that. Radio will always have a space in life. But it’s not innovation.”
Assistants and machine-to-machine marketing
“The concept of digital assistants - the intermediates between you and the world. They follow you, they know everything about you, you grant them your data. The problem is digital assistants become the ones who are going to go to the world, search, get your content, your offers, everything. Whoa! That’s a hell of a difference!
80% of our spend is repetitive. We did some research with Econsultancy in the US
, but I think it’s applicable elsewhere. 80% of people say it would be awesome if their technology helped them to find the perfect option.
I can envision a world in which marketing would be machine-to-machine. The new marketing discipline would be how to target digital assistants and negotiate a price for winning with digital assistants. Machine-to-machine marketing is very feasible.
A command with repetitive buying can be ‘buy the soap I bought last week,’ which creates a pressure for marketers. The incumbent is safe. The challenger finds it difficult in that scenario in which people don’t consider brands. How to make your brand considered is a whole new discipline in that scenario. “
Truly creative uses of AI are on the horizon
“We as Microsoft have a principle that I strongly believe. We believe in artificial intelligence that enhances human beings. It’s not about replacing human beings. It’s about empowering human beings. It’s the tools I give you for you to express your creativity; not for artificial intelligence to be more creative than you. We really want artificial intelligence that gives you capabilities that you didn’t have before and now you can do even more, even better. Most things we’ve been applying artificial intelligence for are totally connected to that.
In that sense, the process of evolution is as a creative person you get ignited by the possibilities of the media - the tools. People to know better what is in those tools to realise the possibilities expanding. If I have the possibility of knowing my consumer at that level, what would I do differently? Now my creativity is applied to a new horizon.
It’s also an adjustment for consumers. I remember years ago you had chat on some websites. As soon as I realised it was a machine dialogue I just cut it out because I knew it was going to be dead before starting. Now it’s different. I find myself more often dialoguing with a chatbot and it works much better. So I’m now more engaged with those things than I was five years ago.
Creativity will use this. I am confident.”